Emission Sustainability

The test is deliberately blunt. Put both sides in dollars over the same window and compare. Osmosis in 2023 is the cleanest illustration of what a failing ratio looks like: over the preceding 90 days, daily trading fees on the exchange ran about $20k, with a one-day peak of $80k during the USDC depeg on March 11, against a daily emission of about $343k. Even under the generous assumption that liquidity providers took every fee dollar and used it to buy OSMO back, that would not be enough to maintain the token's price. The gap is not a rounding error to be closed by better execution; it is an order of magnitude. The article names this as the structural problem facing decentralised exchanges that pay incentives to liquidity providers, and the shape of it recurs across every incentive-funded protocol.

The two sides of the ratio have different natures, which is why the test has to be repeated rather than passed once. Contractual vesting is close to fixed: it is written into a vesting contract and arrives regardless of conditions. Issuance rules are not, they are governance variables, and four issuance schedules covered elsewhere in this glossary have already been rewritten by a vote: Cosmos proposal #848, Osmosis OSMO 2.0, Stacks SIP-045 and Monad MIP-12. Fee income is a flow that rises and falls with usage. Buybacks funded by that flow are therefore flow-dependent rather than structural, and when they run smaller than scheduled emission they fail to meaningfully alter the net supply trajectory, functioning as short-term demand injections rather than long-term supply reduction. A protocol can pass the test in a busy quarter and fail it in a quiet one without changing a single parameter.

Running the test properly means being strict about what counts on the deflation side. A forward reading takes the next twelve months of scheduled unlocks and subtracts only the recurring supply cuts a protocol actually delivers, measured from trailing activity and stripped of one-time events, because future burns are not scheduled anywhere. Only mechanisms that destroy supply count, so a buyback that holds the tokens in a trust and a buyback that redistributes them to stakers both score zero against emission no matter how much cash they deploy. Applied to eleven programs with clean on-chain data as of mid-2026, only BNB came out cleanly net-deflationary. RAY is the case that looks like it should and does not: its emissions run only about 1.9M RAY a year, far below the pace of its fee-funded buyback, so the buyback takes roughly 6.8% of supply off the market over a year, but Raydium's documentation says bought-back RAY is held by the protocol at a public on-chain address rather than burned. That is supply parked, not retired, and by the rule above it scores zero against emission.

Two cautions keep the test honest. First, offsetting meaningfully is not the same as netting negative. Hyperliquid runs the largest fee-funded buyback in the market and closes it into a burn, and on Tokenomist's own calculation it is still net inflationary at roughly 47% a year, because a recurring burn near 14.3M HYPE sits against roughly 119M of scheduled unlocks. Second, a failing ratio today is not automatically a verdict. Research from Artemis finds buybacks are most effective only after clear product-market fit, and heavy early emission can be a deliberate bootstrap; the real question is whether the fee base grows into the schedule before the schedule runs out. Sometimes it does not. Kaito's buyback appears to have paused around September 2025, a date resting on secondary reporting because Kaito never announced it, and its Yaps product was sunset in January 2026. Its net emission now reads near 100% of circulating supply a year, a figure inflated by the fact that only about 24% of its supply circulated as of 2026-08-14, rising to about 27% after the unlock on 2026-08-19. A fee-funded defence lasts only as long as the fee base behind it.